# Lab 3 Exercise Key

# 1. Which model comparison?
# Augmented: Wk4IAT = b0 + b1*Trainhours + e
# Compact:   Wk4IAT = bo + e

# We care about the test of b1.

# 2. lm
mEx = lm(Wk4IAT ~ 1 + Trainhours, data=d)

# 3.
modelSummary(mEx, t=F)

# 4. 
# a. People who have spent 0 hours at diversity trainings have, on average, a non-zero IAT 
#    score at week 4 (M = .541).
# b. Every one hour increase of time spent in diversity training predicted a .04 point decrease
#    in week 4 IAT scores.
# c. Both the parameters are statistically significant (have less than a 5% probability of occurring
#    by chance if the null hypothesis is true).
# d. Yep!

# 5. 
modelEffectSizes(mEx)
# peta2 = .0592. Training hours explains about 6% of the variance in week 4 IAT scores.

# 6. 
d$TrainhoursC = d$Trainhours - mean(d$Trainhours)
mExC = lm(Wk4IAT ~ 1 + TrainhoursC, data=d)

# 7. 
modelSummary(mExC, t=F)
# a. Only the estimate for the intercept parameter.
# b. b1 is the same. b0 now represents the predicted week 4 IAT score for someone who has spent
#    an average number of hours in diversity training.
# c and d
Cs = data.frame(Trainhours = c(2,6))
modelPredictions(mEx, Cs)
# .527 for someone who has attended 2 hours, .357 for someone who has attended 6 hours.

##################
#### GRAPHING ####
##################

plotEx = ggplot(d, aes(x = Trainhours, y = Wk4IAT))
plotEx
plotEx = plotEx + geom_point(position=position_jitter(.1))
plotEx
pY = data.frame(Trainhours = seq(min(d$Trainhours),max(d$Trainhours), length=100))
pY = modelPredictions(mEx, pY)
plotEx = plotEx + geom_smooth(data=pY, aes(ymin=CILo, ymax=CIHi, x = Trainhours, y = Predicted), 
                              stat='identity', color='green')
plotEx
plotEx = plotEx + labs(x = "Hours of diversity training", y= "IAT scores at week 4") + theme_bw(base_size = 14)
plotEx

# The researchers hypothesized that diversity training lowers IAT scores. Participants indicated how
# many hours they had spent at diversity trainings and then, 4 weeks later, completed an IAT. The
# more time participants spent at diversity training, the lower their implicit bias as measured by
# the IAT, F(1,76) = 4.78, p = .032. This variable explained 5.9% of the variance in IAT scores.